Distinguish between simple random sampling and complex random sampling design.
Answers
ANSWER__✌️
There’s no single thing called “complex sampling” or “complex random sampling”. There are a variety of sampling methods.
One set of methods involves stratification: That is, we define various group (strata) within the population and sample within each group. Stratification works well when there are strata that are homogenuous with regard to the thing being investigated, but a lot of variation across strata. So, for example, if we were testing the effects of a new drug and we knew that the drug was likely to act differently in certain groups of people (say, people with some other medical condition) we might stratify on that other condition.
Another set involves clustering: This means that we define clusters within a population (usually a lot of clusters) and then choose some clusters randomly and sample within those clusters (sometimes, a census is attempted in each chosen cluster). Often, clusters are geographic and cluster sampling makes it a lot easier to gather data. For instance, you might divide a region into small tracts and sample within certain tracts and not others, thus saving a lot of travel time and expense.
Note that in stratified sampling there are only a few strata and we sample in each of them, while in cluster sampling there are many clusters and we sample in only some of them.
There are variations on each of these.
There are also non-random sampling methods, some of which are useful in some situations. For instance, rather than try to get a random sample of a patient population, we might take every 10th person to come to a hospital with a certain complaint. This is close to random, but not quite. Random digit dialing is not quite a random sample (because some people have no phone and some have more than one).
In some cases, we have to go quite far from random. Sometimes there is simply no list of all the people (or other subjects) that we want to sample. For instance, there is no list of all the gay people in the USA. Indeed, trying to get a random sample of gay people (even in a place where homosexuality is not highly stigmatized) is notoriously difficult.
It’s hard to sample people who engage in stigmatized or illegal activity. Here, snowball sampling is often useful. In snowball sampling, we find a few people who engage in the activity and then ask them to suggest other people. This also works with rare populations, even if not stigmatized, as long as there are fairly strong network ties.
HOPE IT HELPS YOU ✌✌